Impaired aerobic capacity and premature fatigue preceding muscle weakness in the skeletal muscle Tfam-knockout mouse model

ABSTRACT Mitochondrial diseases are genetic disorders that lead to impaired mitochondrial function, resulting in exercise intolerance and muscle weakness. In patients, muscle fatigue due to defects in mitochondrial oxidative capacities commonly precedes muscle weakness. In mice, deletion of the fast-twitch skeletal muscle-specific Tfam gene (Tfam KO) leads to a deficit in respiratory chain activity, severe muscle weakness and early death. Here, we performed a time-course study of mitochondrial and muscular dysfunctions in 11- and 14-week-old Tfam KO mice, i.e. before and when mice are about to enter the terminal stage, respectively. Although force in the unfatigued state was reduced in Tfam KO mice compared to control littermates (wild type) only at 14 weeks, during repeated submaximal contractions fatigue was faster at both ages. During fatiguing stimulation, total phosphocreatine breakdown was larger in Tfam KO muscle than in wild-type muscle at both ages, whereas phosphocreatine consumption was faster only at 14 weeks. In conclusion, the Tfam KO mouse model represents a reliable model of lethal mitochondrial myopathy in which impaired mitochondrial energy production and premature fatigue occur before muscle weakness and early death.


INTRODUCTION
Mitochondrial diseases are caused by genetic mutations in mitochondrial or nuclear DNA which affect the function of the mitochondrial respiratory chain thereby leading to a defective oxidative phosphorylation (Ishikawa and Nakada, 2020;Parikh et al., 2015;Thorburn, 2004). Skeletal muscles are commonly affected given that they require a high amount of energy and are highly dependent on oxidative phosphorylation for energy production. Clinical manifestations of mitochondrial myopathy generally include exercise Disease Models & Mechanisms • DMM • Accepted manuscript intolerance, muscle weakness, and muscle wasting (Petty et al., 1986). A main histological hallmark of mitochondrial myopathies is the presence of ragged-red fibers, which are muscle fibers exhibiting mitochondrial aggregates in subsarcolemmal regions (Milone and Wong, 2013).
The mitochondrial oxidative phosphorylation system consists of five multi-heteromeric complexes embedded in the inner mitochondrial membrane and composed of subunits encoded by both nuclear and mitochondrial DNA (except complex II which is only encoded by nuclear DNA) (Larsson and Oldfors, 2001;Rotig and Munnich, 2003). The mitochondrial transcription factor A (Tfam) encoded by nuclear DNA is a protein with a key role in mitochondrial DNA maintenance (Larsson et al., 1998). Disruption of Tfam has been shown to cause a global deficiency of all mitochondrial DNA-encoded proteins, hence affecting complexes I, III, IV, and V (Wredenberg et al., 2002).
The mouse model with a selective Tfam knockout in fast-twitch skeletal muscle fibers (Tfam KO) has been shown to mimic pathological features typically reported in patients with mitochondrial myopathies, including ragged-red fibers with abnormally positioned and enlarged mitochondria, atrophic fibers, COX-deficient fibers, and reduction of the respiratory chain enzyme activities (Aydin et al., 2009;Wredenberg et al., 2002). The Tfam KO mice do not live more than ~20 weeks and appear healthy until 12-14 weeks when a progressive clinical phenotype becomes evident (Gineste et al., 2015;Wredenberg et al., 2002).
Experiments conducted in isolated whole muscles and intact single muscle fibers of Tfam KO mice aged of 12-14 weeks have shown muscle weakness (i.e., lower muscle force production) associated with a prolonged and excessive mitochondrial calcium (Ca 2+ ) accumulation (Aydin et al., 2009;Gineste et al., 2015). However, muscle fatigue (i.e., drop in force during sustained or repeated contractions) remained unaffected and has been explained by an increased proportion of the muscle fiber volume being occupied by mitochondria combined with a lower energy expenditure during contractions. This reduced cost would be due to a reduced sarcoplasmic reticulum (SR) calcium (Ca 2+ ) release resulting in a decreased activation of the energy-consuming cross-bridges and SR Ca 2+ pumps (Aydin Disease Models & Mechanisms • DMM • Accepted manuscript et al., 2009;Wredenberg et al., 2002). A similar lack of effect on muscle fatigue has been previously demonstrated in nemaline myopathy mouse models that also display a severe muscle weakness (Gineste et al., 2013;Gineste et al., 2020) and might then be due to the fact that experiments have been conducted in relatively late stages of the disease, i.e., when severe muscle weakness was established and the resulting energy consumption during contractions was decreased. Accordingly, premature fatigue (i.e., faster drop in force) related to decreased mitochondrial oxidative capacity has been shown to precede overt muscle weakness in mouse models with mitochondrial dysfunction as well as in patients with mitochondrial myopathies (Finsterer, 2020;Tarnopolsky, 2016;Yamada et al., 2012).
In the present study, we hypothesized that faster muscle fatigue and impaired mitochondrial energy production precede muscle weakness in the Tfam KO mouse model. In vivo metabolic changes, force production and muscle fatigue were assessed in Tfam KO mice at two different time-points, i.e., before mice enter the stage with obvious symptoms (i.e., at 11 weeks) and when mice are about to enter the terminal stage with progressive muscle weakness (i.e., at 14 weeks). These in vivo investigations of the skeletal muscle function were combined with in vitro experiments on isolated muscles to study whether there were structural alterations indicating increased mitochondrial Ca 2+ influx via mitochondriaassociated membranes (MAM) (Csordas et al., 2018), which might explain the excessive mitochondrial Ca 2+ uptake observed in Tfam KO muscle fibers during repeated contractions (Aydin et al., 2009;Gineste et al., 2015).

Progressive body weight loss is associated with skeletal muscle atrophy without fatty replacement of muscle in Tfam KO mice
Skeletal muscle atrophy and replacement of skeletal muscle tissue by fat deposition or fibrotic tissue is a frequent feature in muscular disorders and has been reported in patients with mitochondrial myopathy, although it is not a common feature (Olsen et al., 2003; Disease Models & Mechanisms • DMM • Accepted manuscript Theodorou et al., 2012). Therefore, we measured body weight and hindlimb muscles volume using magnetic resonance imaging (MRI) in Tfam KO mice and their control littermates, which will be referred to as WT. We also determined whether muscle was replaced by fatty infiltration.
At 11 weeks of age, Tfam KO and WT mice had a similar body weight while at 14 weeks body weight was ~15% lower in Tfam KO than in WT mice (Fig. 1A). While no difference was observed between Tfam KO mice between the two ages, body weight was higher in WT mice at 14 weeks compared to 11 weeks.
The volume of hindlimb muscles was ~15% lower in Tfam KO than in WT mice at 14 weeks while no difference existed at 11 weeks ( Fig. 1B and 1C), indicating muscle atrophy at 14 weeks in Tfam KO mice. No difference was observed between 11 weeks and 14 weeks for the two genotypes.
Neither intramuscular fatty infiltration nor fibrotic tissue was detected in Tfam KO mice at either age as illustrated by the absence pixels with higher intensities in the muscle region ( Fig. 1C).

Reduced maximal force at 14 weeks in Tfam KO mice is related to muscle atrophy
Patients with mitochondrial myopathies usually present skeletal muscle weakness (i.e., decreased force production) (Moggio et al., 2014), which is a common feature reported in neuromuscular diseases (Ahmed et al., 2018;Scelsi, 1992) and which may be due to skeletal muscle atrophy and/or impaired contractile function within the muscles. In order to determine whether the Tfam KO mice showed skeletal muscle weakness, we measured maximal force production (at 150 Hz) and submaximal force production (at 20 Hz that gives <50% of maximal force) in the unfatigued state. We assessed both absolute force and specific force, i.e., force normalized to hindlimb muscle volume to assess contractile function independently of hindlimb muscle size.

Disease Models & Mechanisms • DMM • Accepted manuscript
Absolute maximal tetanic forces did not differ between Tfam KO and WT mice at 11 weeks whereas force was ~15% lower in Tfam KO mice at 14 weeks ( Fig. 2A). When normalized to hindlimb muscles volume, force was similar in Tfam KO and WT mice at both ages, indicating that muscle weakness at 14 weeks in Tfam KO mice is related to muscle atrophy. No difference in maximal force was observed between 11 weeks and 14 weeks for the two genotypes. No significant difference was found for absolute and specific submaximal forces between groups for the same age. Nevertheless, both absolute and specific force at 20 Hz increased for WT mice between 11 and 14 weeks while no significant change was observed for Tfam KO mice (Fig. S1).

Faster fatigue in Tfam KO than in WT mice at 11 and 14 weeks
Premature muscle fatigue (i.e., faster decline in force during sustained or repeated contractions) is a typical feature in patients with mitochondrial myopathy (Finsterer, 2020;Tarnopolsky, 2016). We measured force production during 80 repeated submaximal contractions (1.5 s contractions given every 7.5 s). Specific force was significantly decreased in Tfam KO mice from the 15 th to 50 th contraction at 14 weeks, whereas it was similar between the two groups at 11 weeks ( Fig. 2B and 2C). Absolute force was significantly lower in Tfam KO than in WT mice from 25 th to 35 th contractions at 11 weeks and from 10 th to 50 th at 14 weeks (Fig. S2).
We performed two additional measurements to characterize changes in force production during fatiguing stimulation. First, the total specific force produced throughout the exercise period was similar between WT and Tfam KO mice at 11 weeks, while it was significantly reduced by ~35% at 14 weeks in Tfam KO mice as compared to WT mice (Fig.   2C). We also measured the contraction that gives 50% of the initial force (C 50 ) because the rate of decline in force differed largely in the initial part of fatiguing stimulation between Tfam KO and WT mice. The results revealed a markedly faster initial force decline (lower C 50 ) in Tfam KO than in WT mice at both ages ( Fig. 2D) Fig. 2B). One likely explanation for this is that towards the end of the stimulation period, force was maintained primarily by highly fatigue resistant slow-twitch muscle fibers, which are not genetically afflicted in the fast-twitch muscle fiber-specific Tfam KO mouse.

The muscle energetic status at rest is affected in Tfam KO mice at both ages
Patients with mitochondrial myopathy present impaired muscle metabolism at rest and particularly during exercise (Argov et al., 1987;Matthews et al., 1991;Radda et al., 1985;Taylor et al., 1994). Thus, we used phosphorus-31 magnetic resonance spectroscopy ( 31 P-MRS) to study differences in energy metabolites at rest and during repeated contractions ( Fig. 3A). At rest, [phosphocreatine] to [ATP] ratio was not significantly different between Tfam KO and WT mice at 11 weeks but significantly lower at 14 weeks (Fig. 3B). Resting [inorganic phosphate] was significantly higher in Tfam KO than in WT mice at both ages and this metabolic impairment in Tfam KO mice progressed between 11 and 14 weeks (Fig. 3C).
Resting pH i was not significantly different between groups at either 11 or 14 weeks (Fig. 3D).
The increase in [inorganic phosphate] together with the reduced phosphocreatine] to [ATP] ratio at 14 weeks in Tfam KO mice indicate an impaired metabolic regulation with inadequate matching between oxidative ATP supply and demand in the resting state.  (Fig. 4A). The initial rate of phosphocreatine consumption was similar in Tfam KO and WT mice at 11 weeks, whereas it was ~70% higher in Tfam KO mice than in WT mice at 14 weeks (Fig. 4B). The magnitude of [phosphocreatine] depletion was significantly larger in the Tfam KO mice than in the WT mice at both 11 and 14 weeks (Fig. 4C). The increase in [inorganic phosphate] evolved as a mirror of [phosphocreatine], i.e., inorganic phosphate levels increased rapidly during the first 2 minutes of exercise followed by a plateau from 5 min to the end of the stimulation period ( Fig. 4D & 4E). An increased cytosolic inorganic phosphate levels are considered to be a major cause of force decline in skeletal muscle fatigue (Westerblad and Allen, 1996). Having this in mind, we plotted the relation between force and [inorganic phosphate] after 2.5 min of fatiguing stimulation (i.e., covering the part of fatiguing stimulation where both these parameters showed rapid changes) for individual mice at 11 and 14 weeks ( Fig. 4F) and found a strong significant negative correlation between these two variables (r=-0.69; P<0.001).
The decrease in pH i during the fatiguing stimulation was similar in Tfam KO and WT mice with the only exception being a significant less acidosis at 5 min of stimulation in Tfam KO than in WT mice at 14 weeks (Fig. 4G). Decreased pH i is commonly considered as a major cause of force decline in fatigued skeletal muscle, although its role is currently debated (Westerblad, 2016). Plotting the relation between force and pH i after 2 min of fatiguing stimulation did not show any significant correlation (r=-0.01; P=0.93; Fig. 4H).

Slowed recovery kinetics of [phosphocreatine] in Tfam KO mice at 11 and 14 weeks
Phosphocreatine recovery kinetics are related to the end-exercise rate of oxidative ATP synthesis and a slowed phosphocreatine recovery has been recognized as indicative of an impaired mitochondrial oxidative function (Arnold et al., 1984), and a slower rate has been commonly reported in patients with mitochondrial myopathies (Barbiroli et al., 1997;Bendahan et al., 1992;Taylor et al., 1994). We evaluated the recovery time-courses of both phosphocreatine and inorganic phosphate for 10 min after the end of the fatiguing protocol (see Fig. 4A and 4B). The initial rate of phosphocreatine recovery was significantly lower in

Disease Models & Mechanisms • DMM • Accepted manuscript
Tfam KO than in WT mice at both 11 and 14 weeks (Fig. 5A). Likewise, at both ages the time constant of phosphocreatine recovery (Ƭ PCr ) was significantly longer in the Tfam KO than in WT mice at both ages ( Fig. 5B) and [phosphocreatine] was still ~15% lower in Tfam KO than in WT mice at the end of the 10 min recovery period (Fig. 5C, Table S1). Additionally, [phosphocreatine] was not fully returned to resting levels after the 10min of recovery in Tfam KO mice (Fig. 4A, Table S1). The decrease in [inorganic phosphate] during recovery followed the same time course and was significantly higher in Tfam KO than in WT mice at the end of the recovery period at both ages (Fig. 5D). Moreover, [inorganic phosphate] at the end of the recovery period was significantly higher than the resting level in Tfam KO mice (Fig. 4D, Table S1). During the post-stimulation recovery period, pH i gradually increased towards its resting level at a similar rate in Tfam KO and WT mice (Fig. 4G).

The contact points at the MAM level are decreased in Tfam KO muscle
To study whether alterations in MAM might be involved in the excessive mitochondrial Ca 2+ uptake in Tfam KO muscle fibers (Aydin et al., 2009;Gineste et al., 2015), mitochondria and MAM were isolated from gastrocnemius muscles of 14 weeks WT and Tfam KO mice.
No statistically significant difference between WT and Tfam KO mice was observed for the MAM protein content related to muscle weight ( Fig. 6A) or isolated mitochondria (Fig. 6B), or for total mitochondrial protein content related to muscle weight (Fig. 6C). The physical SRmitochondria interaction, assessed as the in situ protein-protein interaction between the SR Ca 2+ channel inositol 1,3,4-triphosphate receptor 1 (IP 3 R1) and the mitochondrial outer membrane voltage-dependent anion channel 1 (VDAC1) (Szabadkai et al., 2006), was significantly lower in Tfam KO than in WT mice (Fig. 6D).

DISCUSSION
In this study, we used a mouse model of lethal mitochondrial myopathy, the fast-twitch skeletal muscle-specific Tfam KO mouse. Experiments were performed at an early state of the disease process and our results show an impaired mitochondrial energy production accompanied by a faster force decline during repeated contractions in Tfam KO than in WT mice; these impairments occurred before any signs of reduced hindlimb muscle volume and decreased force production in the unfatigued state. In other words, muscles of Tfam KO mice showed an increased muscle fatigue before entering the state with progressive muscle atrophy and muscle weakness. These results extend those from previous studies showing a severe muscle weakness in more advanced stages of the disease (Gineste et al., 2015;Wredenberg et al., 2002) and support those reported in patients related to a progression from early premature fatigue to disabling muscle weakness already at rest (Finsterer, 2020;Tarnopolsky, 2016).
Contractile function in the unfatigued state was impaired in Tfam KO mice only at 14 weeks and at maximal force and was directly related to the skeletal muscle atrophy occurring at this age as normalized force was not different from that in WT mice. The absence of muscle weakness independently from muscle atrophy at maximal and submaximal forces contrasts with previous results showing a ~20% reduction in the force produced by isolated single fast-twitch muscle fibers of Tfam KO mice tested at a similar stage of the disease process, which was related to decreased SR Ca 2+ release (Aydin et al., 2009). This apparent difference might be due to the fact that in vivo the decreased muscle fiber force production might be compensated by various non-contractile structures contributing to force transmission which are partly missing in experiments on isolated single fibers (Cornwall, 1984;Ward et al., 2020). A decreased force production in fast-twitch muscle fibers of Tfam KO mice might also be masked in the present in vivo experiments by a substantial proportion of the muscle being occupied by normally contracting slow-twitch fibers in which Tfam was not knocked out (Wredenberg et al., 2002). In this context it can be noted that the maintained low force levels observed towards the end of fatiguing stimulation likely reflect sustained Disease Models & Mechanisms • DMM • Accepted manuscript force production in fatigue resistant slow-twitch fibers. Accordingly, these maintained low force levels were of similar size in Tfam KO and WT mice, which fits with the normal genetic status of slow-twitch fibers in our Tfam KO mouse model.
Conversely, muscle fatigue induced by repeated submaximal contractions developed markedly faster at both 11 and 14 weeks in Tfam KO mice when compared to WT mice, but was associated with reduced force production in the unfatigued state only at 14 weeks. This result highlights that accelerated muscle fatigue precedes muscle weakness in Tfam KO mice. The altered muscle fatigue was particularly marked in the initial part of fatiguing stimulation. This early undue fatigue in the Tfam KO mice clearly mirrors the exercise intolerance commonly observed in patients with mitochondrial myopathy (Taivassalo and Haller, 2005;Taivassalo et al., 2003). Accelerated fatigue during repeated tetanic stimulation was not previously reported in single muscle fibers isolated from Tfam KO mice at a similar stage in the disease process (Aydin et al., 2009). Three differences between the present and the previous study might explain this discrepancy. First, tetanic force and free cytosolic [Ca 2+ ] ([Ca 2+ ] i ) were already reduced at the start of fatiguing stimulation in the previous single fiber experiments. This means that less energy was used for cross-bridge cycling and SR Ca 2+ reuptake during contractions and hence the stress on mitochondrial energy production was lowered, which would slow fatigue development (Cheng et al., 2019). Second, fatigue was induced by repeated submaximal (~50% of the maximal force) contractions in the present study, whereas close to maximal (~85% of the maximal force) contractions were used in the

Disease Models & Mechanisms • DMM • Accepted manuscript
Our results show clear-cut signs of impaired intracellular energy metabolism before, during and after the fatiguing stimulation bout in Tfam KO muscles at 11 and 14 weeks, which is consistent with the impaired mitochondrial oxidative metabolism observed in mitochondrial myopathy patients (Arnold et al., 1985;Hayes et al., 1985;Matthews et al., 1991;Moller et al., 2005;Taylor et al., 1994). In the resting state, Tfam KO muscles displayed a progressive increase in [inorganic phosphate], which would be required to maintain an adequate rate of mitochondrial ATP synthesis as mitochondrial function declines (Wu et al., 2007).
During the fatiguing protocol, phosphocreatine breakdown was larger in Tfam KO mice at both ages and faster in the initial part of the exercise at 14 weeks. These changes in [phosphocreatine] were mirrored by an increase in [inorganic phosphate] and indicate an increased dependency on phosphocreatine hydrolysis to maintain ATP levels during repeated contractions. Moreover, we observed a good correlation between the increase in [inorganic phosphate] and the decrease in force during the stimulation period (see Fig. 4F), which fits with inorganic phosphate being a major causative factor in acute muscle fatigue that afflicts cross-bridge force production, myofibrillar Ca 2+ sensitivity and SR Ca 2+ release (Allen et al., 2008).
In addition to phosphocreatine breakdown, the major anaerobic energy source is glycolysis ending with formation of lactate and H + when coupled to ATP hydrolysis. Intramuscular acidosis has not been reported as a distinctive characteristic of mitochondrial diseases (Arnold et al., 1985;Hayes et al., 1985;Taylor et al., 1994). Accordingly, resting pH i was unaltered in Tfam KO mice at both ages and so in agreement with reports from patients with mitochondrial myopathies (Arnold et al., 1985;Hayes et al., 1985;Moller et al., 2005).
Moreover, we observed no difference in pH i changes during exercise and recovery periods between Tfam KO and WT mice except for a slightly larger acidification at 5 min of stimulation in WT mice at 14 weeks. These results imply that the faster fatigue in Tfam KO muscle is independent of acidosis. This implication is emphasized by the lack of correlation between decreased pH i and reduced force during fatiguing stimulation (see Fig. 4H) and fits

Disease Models & Mechanisms • DMM • Accepted manuscript
with previous results demonstrating that acidosis is not the primary cause of exercise intolerance in patients with mitochondrial myopathy (Vissing et al., 2001).
Finally, a larger PCr consumption together with a similar or a lower force production (11 weeks and 14 weeks, respectively) might reflect a higher energy cost of contraction. It can be noted that during the first minute of fatiguing stimulation, where ATP production relies on anaerobic processes in both Tfam KO and WT muscles (Lannergren and Westerblad, 1991), the rate of force decline was similar in Tfam KO and WT muscles (see Fig. 4B). Thereafter, when aerobic ATP production becomes increasingly more important, fatigue developed at a faster rate in Tfam KO than in WT muscles. These results imply that the accelerated fatigue development in Tfam KO muscle is due to impaired mitochondrial oxidative capacity rather than increased energy cost of contraction, but further experiments directly addressing this issue are needed to provide a solid answer to this question.
The post-stimulation recovery kinetics of [phosphocreatine] can be used as an index of the mitochondrial oxidative capacity in vivo (Arnold et al., 1984;Lanza et al., 2011). The rate of [phosphocreatine] recovery was markedly decreased and the 10 min recovery period was not long enough for a complete recovery Tfam KO mice, which is in agreement with the slow [phosphocreatine] recovery kinetics commonly reported in patients with mitochondrial myopathies (Barbiroli et al., 1997;Bendahan et al., 1992;Kornblum et al., 2005;Moller et al., 2005;Radda et al., 1982;Taylor et al., 1994). Intriguingly, the slowing of [phosphocreatine] recovery was similar in 11 and 14 weeks Tfam KO mice thereby suggesting that the decline in mitochondrial oxidative capacity was not progressing within this time frame. This is in Excessive Ca 2+ accumulation in mitochondria during contraction has been suggested to be a main causative factor of muscle weakness in Tfam KO mice (Aydin et al., 2009) and the SR-mitochondria interface has been proposed to be the Ca 2+ entry site (Gineste et al., 2015). Indeed, MAM can allow Ca 2+ fluxes from the SR to mitochondria and thereby speed up mitochondrial respiration (Csordas et al., 2018 Hernandez-Ochoa and Schneider, 2018;Santulli et al., 2017;Shishmarev, 2020).
Furthermore, on the mitochondrial side, the MAM-related matrix protein cyclophilin D (also called peptidyl-prolyl cis-trans isomerase F) is involved in the excessive Ca 2+ uptake as illustrated by results showing (i) partial inhibition of the uptake by application of the cyclophilin D-binding inhibitor cyclosporine A, (ii) an increased protein expression of cyclophilin D in skeletal muscle from both Tfam KO mice and mitochondrial myopathy patients, and (iii) an increased life-span of cyclosporine A-treated Tfam KO mice (Aydin et al., 2009;Gineste et al., 2015). Thus, the contraction-mediated aberrant mitochondrial Ca 2+ accumulation in Tfam KO muscle is unlikely to occur via VDAC-IP 3 R1-dependent interactions, but a role of MAM that involves ryanodine receptor 1-cyclophilin D interactions cannot be excluded.
Although the present Tfam KO mouse model mimics major features of human mitochondrial myopathies, it also has limitations. For instance, Tfam is knocked out only in fast-twitch muscle fibers so that there are still intact slow-twitch fibers that may disguise functional impairments and this differs from patients with mitochondrial myopathy where

Disease Models & Mechanisms • DMM • Accepted manuscript
mutations affect all muscle fiber types. Moreover, mitochondrial mutations in humans will be present in various cell types and functional impairments can be seen in other organs than skeletal muscle. Finally, Tfam KO mice show a rapidly progressing myopathy with only a few weeks between the first overt symptoms and death, which makes investigations in the later stage of the disease cumbersome.
In conclusion, disorders related to an impaired mitochondrial function are highly diverse and complex with different organs being affected in a rather unpredictable manner.
Numerous mouse models have been generated to better understand the pathogenic mechanisms underlying mitochondrial diseases (Ishikawa and Nakada, 2020). In this study we used fast-twitch muscle fiber-specific Tfam KO mice and showed that this mouse model presents a disease progression similar to that observed in mitochondrial myopathy patients; that is, an initial decline in mitochondrial oxidative capacity accompanied by a faster force decline during repeated contractions (i.e., accelerated development of muscle fatigue), which is followed by a progressive decrease in force production already in the unfatigued state (i.e., muscle weakness), reduction of body weight and premature death. Thus, the present Tfam KO mice provide a good model for studies of mechanisms underlying muscle defects at different stages of mitochondrial myopathies and the knowledge gained can be used to propose novel treatment strategies for these serious diseases where no effective treatment currently exists.

Animals
Tfam KO mice and their control littermates (referred to as WT) were used for experiments. The fast-twitch skeletal muscle fiber-specific Tfam KO mouse model was generated as previously described (Wredenberg et al., 2002). A first group of mice was tested at 11 weeks of age (i.e., before mice enter the stage with severe phenotype) and a second group at 14 weeks of age (i.e., when mice are about to enter the terminal stage with

Study design
Functional, anatomical and metabolic investigations of the left hindlimb muscles of WT and Tfam KO mice were performed strictly non-invasively in vivo. Body weight was measured before mice were subjected to the following protocol. The mechanical performance was assessed by measuring plantar flexor muscles (gastrocnemius [Gas], soleus [Sol], and plantaris) maximal tetanic force, and plantar flexor muscles force production during a fatiguing protocol. Metabolic changes were evaluated before, during, and after the fatiguing stimulation protocol using phosphorus-31 magnetic resonance spectroscopy ( 31 P-MRS). Before the fatiguing protocol, the volume of hindlimb muscles was quantified by magnetic resonance imaging (MRI). Mice were euthanized by cervical dislocation at the end of the force and MR experiments.

Animal preparation
Mice were initially anesthetized in an induction chamber (Equipement vétérinaire; Minerve, Esternay, France) with 4% isoflurane in 33% O 2 (0.5 L/min) and 66% N 2 O (1 L/min) and placed supine in a cradle designed in our laboratory. A custom-built facemask continuously supplied with 1.75% isoflurane in 33% O 2 (0.2 L/min) and 66% N 2 O (0.4 L/min) was incorporated into the cradle and was used to maintain a prolonged anesthesia throughout the experiment. The hindlimb was centered inside a 16 mm-diameter 1 H

Disease Models & Mechanisms • DMM • Accepted manuscript
Helmholtz imaging coil and the belly of the Gas muscle was located above an elliptical (6 x 8 mm) 31 P-MRS surface coil. The foot was positioned on the pedal of a custom-made ergometer with a 90° flexion ankle joint. A force transducer was part of the pedal as previously described (Giannesini et al., 2010). Two rod-shaped surface electrodes integrated into the cradle and connected to an electrical stimulator (Digitimer, Ltd., Model DS7A, Welwyn Springs, UK) were placed on the left hindlimb, one at the heel level and the other one just above the knee joint.

Force output measurements
The analog electrical signal coming out from the force transducer was amplified with a home-built amplifier (Operational amplifier AD620; Analog Devices, Norwood, MA, USA), converted to a digital signal, monitored and recorded on a personal computer using the Powerlab 35/series system (AD Instruments, Oxford, United Kingdom). Skeletal muscle contractions were achieved by non-invasive transcutaneous electrical stimulation elicited with square-wave pulses (0.5 ms duration). The individual maximal stimulation intensity was determined by progressively increasing the stimulus intensity until there was no further increase in peak twitch force. Maximal tetanic force of the plantar flexor muscles was assessed in response to a 150 Hz pulse train (duration = 0.75 s) and submaximal force in response to 20 Hz (duration = 0.75 s). Plantar flexor muscles force production was also measured during a fatigue protocol (80 contractions -40 Hz -1.5 s on, 6 s off).
The peak force of each contraction was measured with the LabChart software (AD Instruments, Oxford, United Kingdom). Regarding the fatigue protocol, the tetanic force was averaged every 5 contractions for the sake of clarity. The total force production was computed as the sum of each individual peak force measured during the fatigue protocol.
The C 50 corresponded to the contraction number at which the force amplitude was 50% of the initial force of the fatiguing exercise. Force was normalized with respect to the volume of hindlimb muscles (see below) to obtain specific force (in mN/mm 3 ).
Anatomical imaging MRI data were acquired at rest, i.e., before fatiguing protocol. Fifteen consecutive contiguous axial slices (thickness = 0.7 mm), covering the region from the knee to the ankle, were selected across the lower hindlimb. Rapid acquisition with relaxation enhancement The border of the whole hindlimb muscle area was manually delineated in the two slices located on the proximal and distal parts. Segmentation of the missing intermediate slices was automatically performed on the basis of image registration procedures as previously described (Ogier et al., 2017). The volume of the hindlimb muscles (mm 3 ) was calculated as the sum of the volume of the ten consecutive largest slices, which were located on the proximal part and covered the largest volume of the hindlimb muscle belly. The first 180 FID were acquired at rest and summed together (n = 1, time resolution = 6 min).
The next 320 FID were acquired during the stimulation period and summed by blocks of 80 during the stimulation procedure (n = 4, time resolution = 160 s). The last 300 FID were acquired during the recovery period and summed by blocks of 50 (n = 6, time resolution = 100 s). The relative concentrations of high-energy phosphate metabolites (phosphocreatine, inorganic phosphate and adenosine triphosphate (ATP)) were obtained by a time-domain fitting routine using the AMARES-MRUI Fortran code and appropriate prior knowledge of the ATP multiplets. As previously suggested (Chance et al., 1981), inorganic phosphate values were normalized to the sum of phosphocreatine and inorganic phosphate values, which were considered as constant in order to take into account potential movement artifacts.
Intracellular pH (pH i ) was calculated from the chemical shift of the inorganic phosphate signal relative to phosphocreatine (Moon and Richards, 1973). The resting [phosphocreatine] to [ATP] ratio was calculated from the peak areas of the phosphocreatine and β-ATP of the spectrum acquired at rest. The initial rate of phosphocreatine degradation (rate of phosphocreatine degradation at the start of exercise; Vi Phosphocreatine degradation ) was calculated as: Vi Phosphocreatine degradation = ΔPhosphocreatine degradation /τPhosphocreatine degradation , where ΔPhosphocreatine degradation is the percentage of phosphocreatine depletion measured at the exercise end (relative to basal value) and τPhosphocreatine degradation is the time constant of phosphocreatine degradation which was determined by fitting the time course of [phosphocreatine] level to a mono-exponential function with a least-mean-squared algorithm.
Similarly, the phosphocreatine recovery kinetic parameters (initial rate of phosphocreatine resynthesis at the start of the recovery period (Vi Phosphocreatine resynthesis ) and [phosphocreatine] Disease Models & Mechanisms • DMM • Accepted manuscript recovery time constant (τ Phosphocreatine ) were determined during the post exercise period by fitting the time course of phosphocreatine resynthesis to a mono-exponential function.
The tissues were finely minced with scissors and the pieces were then homogenized by applying 10-20 strokes in a glass-Teflon tissue grinder. The homogenates were centrifuged at 740 g for 5 min, and the supernatants were taken and similarly centrifuged. The new supernatants were collected and centrifuged at 9,000 g for 10 min at 4°C. The pellets were was added on top of the mitochondrial suspensions to fill up the tubes. The latter were centrifuged at 95,000 g for 30 min at 4°C. The MAM, identified as a round white layer in the middle of the tube, were collected and washed by MRB through centrifugation at 6,300 g /10 min/4 °C and then re-centrifuged in polycarbonate tubes using a 100,000 g/1h/4 °C ultracentrifugation. The ring of MAM at the bottom of the tubes was finally collected. Protein concentration was quantified by the Lowry method.

Disease Models & Mechanisms • DMM • Accepted manuscript
In Situ Proximity Ligation Assay VDAC1 (ab14734) and IP 3 R1 (sc28614) proximity were measured by in situ proximity ligation assay (PLA) (Sigma Duolink kits anti-rabbit plus (DUO92002) and anti-Mouse minus (DUO92004) to detect and quantify SR-mitochondria interactions, as previously described and thoroughly validated (Tubbs and Rieusset, 2016;Tubbs et al., 2014). In muscle tissue, in situ PLA were performed on paraffin-embedded sections, after an antigen retrieval at pH 6, using a bright-field revelation. Dots were quantified using Image J. Results were expressed as number of dots per field. A minimum of 10 images per muscle were analyzed.

Statistical analysis
Data are presented as mean ± SEM. Statistical analyses were performed with GraphPad Prism software version 7.04. Normality was checked using a Shapiro-Wilk test.
Parametric tests were performed when data were normally distributed. If not, non-parametric tests were used. Student's unpaired t-test or unpaired Mann-Whitney test was used to compare VDAC-IP3R1 interaction and MAM, respectively. Two-way ANOVA (genotype x age) was used to compare WT-11w, Tfam KO-11w mice, WT-14w and Tfam KO-14w mice.
Two-way ANOVA with repeated measures on time or contraction number was used to compare force production or metabolic changes during induction of fatigue and recovery in the four groups. When a significant interaction was found, Tukey's post-hoc test was performed. When a main effect was found, Sidak's post-hoc test was performed. Statistical significance was accepted when P<0.05. For the sake of clarity, when an interaction was found, we only show significant differences between: WT and Tfam KO mice at either 11 or 14 weeks; 11 weeks and 14 weeks for either WT or Tfam KO mice (i.e., statistical differences between WT at 11 weeks vs. Tfam KO at 14 weeks and WT at 14 weeks vs. Tfam KO at 11 weeks are not shown).